Man Files Lawsuit Against OpenAI, Alleging Bad Medical Advice
A Florida man filed a lawsuit against OpenAI and CEO Sam Altman, saying that the company’s chatbot, ChatGPT, gave him incorrect medical information that resulted in a near-fatal health crisis. Scott Winters alleges that the chatbot misdiagnosed him with dysautonomia in the lawsuit.
OpenAI And The Lawsuit
The lawsuit, filed by 55-year-old Scott Winters, says he frequently turned to ChatGPT for help when he experienced troubling symptoms. According to the complaint, the chatbot ultimately diagnosed Winters with dysautonomia, a disorder of the autonomic nervous system, and produced a personalized recovery plan with specific instructions for him. On July 13, 2025, according to the suit, Winters suffered a massive pulmonary embolism due to multiple blood clots in both lungs that was nearly fatal.
The alleged diagnosis from ChatGPT was contradicted by the doctors’ notes. Winters’ complaint contends that the chatbot went beyond offering general information and instead provided individualized medical advice that discouraged him from seeking professional care. The suit alleges that, over time, the bot reduced the number of disclaimers in its responses and began offering diagnoses and treatment steps without directing Winters to medical personnel or escalating crisis safeguards.
Medical records cited in the lawsuit, the complaint says, contradict the chatbot’s diagnosis. A clinician’s note included in the filings, the suit says, explicitly rejected dysautonomia as Winters’ condition and described the diagnosis as self-made after internet research. The complaint argues that ChatGPT’s guidance and the alleged failure to insist on professional evaluation contributed to the life-threatening emergency.
Related Tech News
- Foldable Phones Are Finally Going Mainstream, Tech Fans Are Losing It – Total Apex Herald
- Trump Pulls AI Executive Order at Last Minute, Citing U.S. Tech Lead – Total Apex Herald
OpenAI, Health Advice And Broader Concerns
OpenAI has defended the technology while stressing its limits. A company spokesperson told CNET that “ChatGPT is not a doctor and should never be used as a substitute for medical care, diagnosis or treatment. Treating chatbots as the whole story behind people’s medical decisions or outcomes oversimplifies a much bigger challenge and risks getting in the way of people accessing powerful new tools that can aid them in their health journey.” The company also noted that the model Winters used, GPT-4o, has since been retired amid concerns that it was overly agreeable and could cause harm.
Researchers have raised separate concerns about both the accuracy and safety of AI health advice. A February 2026 Nature Medicine study out of Oxford found that while chatbots answered health questions correctly nearly 95% of the time on their own, real people using those same chatbots to work through medical scenarios landed on the correct answer only about 35% of the time — no better than not using a chatbot at all.
A separate Nature Medicine study from Mount Sinai researchers, testing OpenAI’s ChatGPT Health specifically, found the tool under-triaged more than half of cases that physicians classified as true emergencies — telling users it was fine to wait when doctors said they needed immediate care. Together, the studies point to a gap between how AI models perform on medical knowledge benchmarks like the USMLE and how reliably they perform when real people rely on them for real decisions.
Public use of AI for health information is already widespread. OpenAI’s own survey released in January found that three in five U.S. adults had used AI tools for health-related information in the previous three months. That prevalence, advocates say, makes the question of how chatbots present risk and when they should insist on professional care especially urgent.
What’s Next With Winters’ Lawsuit Against OpenAI?
This case puts a human face on a knotty question: when a person uses a chatbot for any help with health, who is responsible if that guidance goes wrong? It forces courts and companies to reckon with how clearly limits are disclosed, how models are built to recognize danger, and what duty developers have when conversational systems start to sound like medical authorities.
OpenAI’s choice to retire GPT-4o shows the company understood the risks, yet the lawsuits make plain that technical fixes alone won’t settle the matter. Families, patients, and clinicians are left anxious about where accountability begins and ends, and judges will soon be asked to weigh harm, intent, and design choices against the very real consequences people say they suffered. For those who nearly lost everything, this is not an abstract policy debate; it is a search for answers, responsibility, and, for some, a measure of justice.
For Winters, the complaint frames the episode as more than a technical failure: it describes a brush with death and a search for accountability after what his lawyers call “life-altering” consequences. The litigation will proceed in the courts, where judges and juries will weigh whether the chatbot’s conduct crossed a line from information to actionable, harmful medical advice.
